1. Memory for General Knowledge and Concepts and Categorization

2. Semantic vs. Episodic Memory
Memories for specific events that you yourself were involved in
Semantic Memory:
General knowledge base
Lots of different ways of organization info, each with different predictions
Tulving proposed LTM should be divided into two memory types
Episodic memory: being able to travel back in time and experience something consciously again, temporally dated (which means that an event happened before, after, or at same time as another event)
Semantic: info arranged according to meaning

3. Memory for General Knowledge
Definitions of words
Verb tenses
Arithmetic facts
Historical facts
Scientific facts
Geography facts
All examples of info/knowledge that is general, not specific to yourself
Not personal, you don’t remember when you learned it, not specific memories
Amnesia example: Gene, damage to frontal/temporal lobes, including left hippocampus
cannot remember events from life, but can remember facts (name of high school)
Another example: woman w/ damage to front temporal lobe, can remember events but not facts (color of a mouse, where to find soup)

4. The Semantic Memory Models
Hierarchical Semantic Network Model
Cognitive economy
Breathes air
Mammal
Live young
pointers
First models of semantic memory build in 70s when information-processing models prominent, and amount of memory available to computers was limited, so researchers concentrating on how to conserve memory space
Think of this and following models as a library, each piece of info is a book, that has a storage place, these places called nodes
One idea: reduce redundant information
Cognitive economy, you store info for fact at highest level of memory possible
Nodes arranged in hierarchies
So, prediction:
if memory organized like this, it should be faster to verify info closer to activated node than info farther from active node
“Bernese mountain dog has an exuberant personality.”
“Bernese mountain dog has live young.”
Dog
Four paws
Node
Tail
Bernese Mountain
Dog
Black, white, rust fur
Exuberant

5. The Semantic Memory Models
Hierarchical Semantic Network Model
Problems with model
Hierarchy?
Animal
If memory was arranged in a hierarchy,
first sentence should be verified faster than second sentence
but that does not happen, second is faster
Mammal
“A pig is a mammal.”
“A pig is an animal.”
Pig

6. The Semantic Memory Models
Hierarchical Semantic Network Model
Problems with model
Typicality effect
“A robin is a bird.”
Bird
Robin
Turkey
If hierarchy,
two sentences should be processed at same time
robin is faster
“A turkey is a bird.”

7. The Semantic Memory Models
Feature Comparison Model
Features
Defining
Characteristic
Memory of a word/concept has a set of ‘features’, a list of features
Two types:
- must be present
- usually present, but not always
This is how it works: “a robin is a bird”
1. feature lists activated (for both words, for both types of features)
2. lists scanned
3. decision: overlap?
4. if not conclusive, defining features only compared more closely
Explains typicality effect (robin has more “bird” features than turkey
Explains fast rejection of a sentence (no overlaps)
Explains category size effect (which is, that a word can be subordinate to another word/concept; faster to confirm a sentence that has words from smaller, lower categories than more abstract ones) – ex. a collie is a dog, a collie is an animal (assume lists being compared, smaller category has more to do with dog than with animal)
Explains “hedges” (something is sort of but not exactly like another thing, bat, bird) (hedge possible b/c using characteristic features but not defining features)
However, arguments that maybe defining features don’t even exist

8. The Semantic Memory Models
Other Network Models
Elaboration of network model
Main diffs:
if one node activated, related nodes activated
similar words close together, and have many connections
connections between nodes can have diff strengths or weights, activation diminished if words less related
no hierarchies, no cognitive economy
However, it is difficult to think of data that would falsify model
Depiction of spreading activation. Once the node for “bread” is excited, the activation travels to related nodes.

9. The ACT model (Adaptive Control of Thought)
The Semantic Memory Models
Other Network Models
Does not use semantic/episodic distinction
Instead, divides memory into three parts (ignore arrows for moment) working memory (info system currently using)
declarative memory (info that can be recalled as words, pictures or symbols, ex. knowledge, facts, info, ideas)
procedural memory (info about actions and consequences of actions) (telephone number example, having to visualize make movements)
Main diff from other models: there are different kinds of nodes (spatial, abstract, procedural)
This model meant to show what happens in general cognition,
The ACT model (Adaptive Control of Thought)

10. The Semantic Memory Models
Other Network Models
The ACT model production rules: - goal - one or more conditions - one or more actions to take
Production rules activated like other nodes, …

11. The Semantic Memory Models
Connectionists Model
Very different from other models
It proposes that there are no “nodes”
Instead there are processing units (neurons), and a fact is encoded as a pattern of activation of these different units
Each unit starts out neutral, with neutral connections b/t units, over time/learning, brain learns that when one unit activated, certain other ones should be as well

12. Schemata
Scripts

13. Implicit vs. Explicit Memory
Process Dissociation Framework

14. Why do we need to know about categories such as “cars,” “people,” “mountains,” and “birds” in order to make sense of our experiences?
Do people in different cultures categorize objects in the same way?

15. Theoretical Descriptions of Concepts
Classical View

16. Theoretical Descriptions of Concepts
Prototype View

17. Theoretical Descriptions of Concepts
Exemplar View

18. Theoretical Descriptions of Concepts
Schemata View

19. Theoretical Descriptions of Concepts
Knowledge-Based View